deploy: 14cdb84

Author: jeremymanning

.buildinfo

@@ -1,4 +1,4 @@
 # Sphinx build info version 1
 # This file hashes the configuration used when building these files. When it is not found, a full rebuild will be done.
-config: 0427e3b1febb1cd3abd56557eed81678
+config: e9a4eab4ea018d73e925b33924d89411
 tags: 645f666f9bcd5a90fca523b33c5a78b7

_sources/assignments/Assignment_1:Hopfield_Networks/README.md

@@ -111,25 +111,16 @@ Create a heatmap:
 
 #### Visualization 2
 
-Plot the expected number of accurately retrieved memories vs. network size.
-
-Let:
-
-- $P[m, N] \in [0, 1]$: proportion of $m$ memories accurately retrieved in a network of size $N$
-- $\mathbb{E}[R_N]$: expected number of successfully retrieved memories
-
-Then:
-
-$$
-\mathbb{E}[R_N] = \sum_{m=1}^{M} m \cdot P[m, N]
-$$
-
-Where $M$ is the maximum number of memories tested.
+Plot the estimated number of accurately retrieved memories as a function of network size.
+You can use the heatmap you made above to estimate the number of accurately retrieved memories:
+  - Choose a target proportion (e.g., $p = 0.8$ or similar)
+  - For each network size you tested to make your heatmap, compute the maximum number of memories for which at least proportion $p$ were successfully retrieved
 
 #### Follow-Up
 
 - What relationship (if any) emerges between network size and capacity?
 - Can you develop rules or intuitions that help predict a network’s capacity?
+- Hint: see page 3 of [Amit et al. (1985)](https://www.dropbox.com/scl/fi/3a3adwqf70afb9kmieezn/AmitEtal85.pdf?rlkey=78fckvuuvk9t3o9fbpjrmn6de)
 
 ---
 
@@ -211,7 +202,7 @@ $$
 Context drift:
 
 - Set $\text{context}^1$ randomly
-- For each subsequent $\text{context}^{t+1}$, copy $\text{context}^t$ and flip ~5% of the bits
+- For each subsequent $\text{context}^{t+1}$, copy $\text{context}^t$ and flip ~10% of the bits
 
 #### Simulation Procedure
 
@@ -221,12 +212,12 @@ Context drift:
    - Set item neurons to 0
    - Run until convergence
    - For each stored memory $j$, compare recovered item to $\text{item}^j$
-   - If ≥99% of bits match, record $j$ as retrieved
+   - If ≥75% of bits match, record $j$ as retrieved
    - Record $\Delta = j - i$ (relative offset)
 
 #### Analysis
 
-- Repeat the procedure (e.g., 100 trials)
+- Repeat the procedure (e.g., 100 trials).  Note that you will need to "reset" the network (i.e., start with an empty weight matrix and re-encode the 10 memories) each time you repeat the simulation.
 - For each $\Delta \in [-9, +9]$, compute:
   - Probability of retrieval
   - 95% confidence interval

_sources/assignments/Assignment_2:Search_of_Associative_Memory_Model/README.md

@@ -1,7 +1,7 @@
 # Assignment 2: Search of Associative Memory (SAM) Model
 
 ## Overview
-In this assignment, you will implement the **Search of Associative Memory (SAM) model** as described in [Atkinson & Shiffrin (1968)](https://www.dropbox.com/scl/fi/rpllozjcv704okckjdy5k/AtkiShif68.pdf?rlkey=i0azhj9mqxws7bxocbl65j88d&dl=1). The SAM model is a probabilistic model of free recall that assumes items are stored in **short-term memory (STS)** and **long-term memory (LTS)**, with retrieval governed by associative processes. You will fit your implementation to [Murdock (1962)](https://www.dropbox.com/scl/fi/k7jc1b6uua4m915maglpl/Murd62.pdf?rlkey=i5nc7lzb2pw8dxc6xc72r5r5i&dl=1) free recall data and evaluate how well the model explains the observed recall patterns.
+In this assignment, you will implement the **Search of Associative Memory (SAM) model** as described in [Kahana (2012), Chapter 7](https://www.dropbox.com/scl/fi/ujl8yvxqzcb1gf32to4zb/Kaha12_SAM_model_excerpt.pdf?rlkey=254wtw4fm7xnpzelno2ykrxzu). The SAM model is a probabilistic model of free recall that assumes items are stored in **short-term memory (STS)** and **long-term memory (LTS)**, with retrieval governed by associative processes. You will fit your implementation to [Murdock (1962)](https://www.dropbox.com/scl/fi/k7jc1b6uua4m915maglpl/Murd62.pdf?rlkey=i5nc7lzb2pw8dxc6xc72r5r5i&dl=1) free recall data and evaluate how well the model explains the observed recall patterns.
 
 ## Data Format and Preprocessing
 The dataset consists of sequences of recalled items from multiple trials of a free recall experiment. Each row represents a participant’s recall sequence from a single list presentation.
@@ -20,6 +20,8 @@ The dataset consists of sequences of recalled items from multiple trials of a fr
 
 ## Model Description
 
+*NOTE: NEED TO UPDATE THIS!!!  Don't start this assignment quite yet...*
+
 You will implement and fit a **Search of Associative Memory (SAM)** model that consists of:
 
 ### 1. **Encoding Stage: Memory Storage**
@@ -140,7 +142,7 @@ Each plot should include:
   - Identifies strengths and weaknesses of SAM.
 
 ## Submission Instructions
-- Submit a **Google Colaboratory notebook** (or similar) that includes:
+- [Submit](https://canvas.dartmouth.edu/courses/71051/assignments/517354) a **Google Colaboratory notebook** (or similar) that includes:
   - Your **full implementation** of the SAM model.
   - **Markdown cells** explaining your code, methodology, and results.
   - **All required plots** comparing model predictions to observed data.

_sources/outline.md

@@ -22,7 +22,8 @@ Note: papers that report on models we will be implementing in the assignments ar
     - free recall and memory search
     - naturalistic memory tasks
   - Readings:
-    - [Atkinson and Shiffrin (1968)](https://www.dropbox.com/scl/fi/rpllozjcv704okckjdy5k/AtkiShif68.pdf?rlkey=i0azhj9mqxws7bxocbl65j88d)*
+    - [Atkinson and Shiffrin (1968)](https://www.dropbox.com/scl/fi/rpllozjcv704okckjdy5k/AtkiShif68.pdf?rlkey=i0azhj9mqxws7bxocbl65j88d)
+    - [Kahana (2012), excerpt from Chapter 7](https://www.dropbox.com/scl/fi/ujl8yvxqzcb1gf32to4zb/Kaha12_SAM_model_excerpt.pdf?rlkey=254wtw4fm7xnpzelno2ykrxzu)*
     - [Chen et al. (2016)](https://www.dropbox.com/scl/fi/wg6fledn7g88ig5mk3kob/ChenEtal16.pdf?rlkey=9jqu7y2apqv2hrj8qepn4alwa)
     - [Heusser et al. (2021)](https://www.dropbox.com/scl/fi/w7z2yvdfzmhowh5hvg53e/HeusEtal21.pdf?rlkey=omad9klqeiu2kc71w7guc5xxq)
   - Data science primer:

assignments/Assignment_1:Hopfield_Networks/README.html

@@ -579,24 +579,19 @@ <h4>Visualization 1<a class="headerlink" href="#visualization-1" title="Link to
 </section>
 <section id="visualization-2">
 <h4>Visualization 2<a class="headerlink" href="#visualization-2" title="Link to this heading">#</a></h4>
-<p>Plot the expected number of accurately retrieved memories vs. network size.</p>
-<p>Let:</p>
+<p>Plot the estimated number of accurately retrieved memories as a function of network size.
+You can use the heatmap you made above to estimate the number of accurately retrieved memories:</p>
 <ul class="simple">
-<li><p><span class="math notranslate nohighlight">\(P[m, N] \in [0, 1]\)</span>: proportion of <span class="math notranslate nohighlight">\(m\)</span> memories accurately retrieved in a network of size <span class="math notranslate nohighlight">\(N\)</span></p></li>
-<li><p><span class="math notranslate nohighlight">\(\mathbb{E}[R_N]\)</span>: expected number of successfully retrieved memories</p></li>
+<li><p>Choose a target proportion (e.g., <span class="math notranslate nohighlight">\(p = 0.8\)</span> or similar)</p></li>
+<li><p>For each network size you tested to make your heatmap, compute the maximum number of memories for which at least proportion <span class="math notranslate nohighlight">\(p\)</span> were successfully retrieved</p></li>
 </ul>
-<p>Then:</p>
-<div class="math notranslate nohighlight">
-\[
-\mathbb{E}[R_N] = \sum_{m=1}^{M} m \cdot P[m, N]
-\]</div>
-<p>Where <span class="math notranslate nohighlight">\(M\)</span> is the maximum number of memories tested.</p>
 </section>
 <section id="follow-up">
 <h4>Follow-Up<a class="headerlink" href="#follow-up" title="Link to this heading">#</a></h4>
 <ul class="simple">
 <li><p>What relationship (if any) emerges between network size and capacity?</p></li>
 <li><p>Can you develop rules or intuitions that help predict a network’s capacity?</p></li>
+<li><p>Hint: see page 3 of <a class="reference external" href="https://www.dropbox.com/scl/fi/3a3adwqf70afb9kmieezn/AmitEtal85.pdf?rlkey=78fckvuuvk9t3o9fbpjrmn6de">Amit et al. (1985)</a></p></li>
 </ul>
 </section>
 </section>
@@ -702,7 +697,7 @@ <h4>Setup: Item–Context Representation<a class="headerlink" href="#setup-itemc
 <p>Context drift:</p>
 <ul class="simple">
 <li><p>Set <span class="math notranslate nohighlight">\(\text{context}^1\)</span> randomly</p></li>
-<li><p>For each subsequent <span class="math notranslate nohighlight">\(\text{context}^{t+1}\)</span>, copy <span class="math notranslate nohighlight">\(\text{context}^t\)</span> and flip ~5% of the bits</p></li>
+<li><p>For each subsequent <span class="math notranslate nohighlight">\(\text{context}^{t+1}\)</span>, copy <span class="math notranslate nohighlight">\(\text{context}^t\)</span> and flip ~10% of the bits</p></li>
 </ul>
 </section>
 <section id="id4">
@@ -715,7 +710,7 @@ <h4>Simulation Procedure<a class="headerlink" href="#id4" title="Link to this he
 <li><p>Set item neurons to 0</p></li>
 <li><p>Run until convergence</p></li>
 <li><p>For each stored memory <span class="math notranslate nohighlight">\(j\)</span>, compare recovered item to <span class="math notranslate nohighlight">\(\text{item}^j\)</span></p></li>
-<li><p>If ≥99% of bits match, record <span class="math notranslate nohighlight">\(j\)</span> as retrieved</p></li>
+<li><p>If ≥75% of bits match, record <span class="math notranslate nohighlight">\(j\)</span> as retrieved</p></li>
 <li><p>Record <span class="math notranslate nohighlight">\(\Delta = j - i\)</span> (relative offset)</p></li>
 </ul>
 </li>
@@ -724,7 +719,7 @@ <h4>Simulation Procedure<a class="headerlink" href="#id4" title="Link to this he
 <section id="id5">
 <h4>Analysis<a class="headerlink" href="#id5" title="Link to this heading">#</a></h4>
 <ul class="simple">
-<li><p>Repeat the procedure (e.g., 100 trials)</p></li>
+<li><p>Repeat the procedure (e.g., 100 trials).  Note that you will need to “reset” the network (i.e., start with an empty weight matrix and re-encode the 10 memories) each time you repeat the simulation.</p></li>
 <li><p>For each <span class="math notranslate nohighlight">\(\Delta \in [-9, +9]\)</span>, compute:</p>
 <ul>
 <li><p>Probability of retrieval</p></li>

assignments/Assignment_2:Search_of_Associative_Memory_Model/README.html

@@ -467,7 +467,7 @@ <h2> Contents </h2>
 <h1>Assignment 2: Search of Associative Memory (SAM) Model<a class="headerlink" href="#assignment-2-search-of-associative-memory-sam-model" title="Link to this heading">#</a></h1>
 <section id="overview">
 <h2>Overview<a class="headerlink" href="#overview" title="Link to this heading">#</a></h2>
-<p>In this assignment, you will implement the <strong>Search of Associative Memory (SAM) model</strong> as described in <a class="reference external" href="https://www.dropbox.com/scl/fi/rpllozjcv704okckjdy5k/AtkiShif68.pdf?rlkey=i0azhj9mqxws7bxocbl65j88d&amp;amp;dl=1">Atkinson &amp; Shiffrin (1968)</a>. The SAM model is a probabilistic model of free recall that assumes items are stored in <strong>short-term memory (STS)</strong> and <strong>long-term memory (LTS)</strong>, with retrieval governed by associative processes. You will fit your implementation to <a class="reference external" href="https://www.dropbox.com/scl/fi/k7jc1b6uua4m915maglpl/Murd62.pdf?rlkey=i5nc7lzb2pw8dxc6xc72r5r5i&amp;amp;dl=1">Murdock (1962)</a> free recall data and evaluate how well the model explains the observed recall patterns.</p>
+<p>In this assignment, you will implement the <strong>Search of Associative Memory (SAM) model</strong> as described in <a class="reference external" href="https://www.dropbox.com/scl/fi/ujl8yvxqzcb1gf32to4zb/Kaha12_SAM_model_excerpt.pdf?rlkey=254wtw4fm7xnpzelno2ykrxzu">Kahana (2012), Chapter 7</a>. The SAM model is a probabilistic model of free recall that assumes items are stored in <strong>short-term memory (STS)</strong> and <strong>long-term memory (LTS)</strong>, with retrieval governed by associative processes. You will fit your implementation to <a class="reference external" href="https://www.dropbox.com/scl/fi/k7jc1b6uua4m915maglpl/Murd62.pdf?rlkey=i5nc7lzb2pw8dxc6xc72r5r5i&amp;amp;dl=1">Murdock (1962)</a> free recall data and evaluate how well the model explains the observed recall patterns.</p>
 </section>
 <section id="data-format-and-preprocessing">
 <h2>Data Format and Preprocessing<a class="headerlink" href="#data-format-and-preprocessing" title="Link to this heading">#</a></h2>
@@ -487,6 +487,7 @@ <h2>Data Format and Preprocessing<a class="headerlink" href="#data-format-and-pr
 </section>
 <section id="model-description">
 <h2>Model Description<a class="headerlink" href="#model-description" title="Link to this heading">#</a></h2>
+<p><em>NOTE: NEED TO UPDATE THIS!!!  Don’t start this assignment quite yet…</em></p>
 <p>You will implement and fit a <strong>Search of Associative Memory (SAM)</strong> model that consists of:</p>
 <section id="encoding-stage-memory-storage">
 <h3>1. <strong>Encoding Stage: Memory Storage</strong><a class="headerlink" href="#encoding-stage-memory-storage" title="Link to this heading">#</a></h3>
@@ -671,7 +672,7 @@ <h2>Grading Criteria<a class="headerlink" href="#grading-criteria" title="Link t
 <section id="submission-instructions">
 <h2>Submission Instructions<a class="headerlink" href="#submission-instructions" title="Link to this heading">#</a></h2>
 <ul class="simple">
-<li><p>Submit a <strong>Google Colaboratory notebook</strong> (or similar) that includes:</p>
+<li><p><a class="reference external" href="https://canvas.dartmouth.edu/courses/71051/assignments/517354">Submit</a> a <strong>Google Colaboratory notebook</strong> (or similar) that includes:</p>
 <ul>
 <li><p>Your <strong>full implementation</strong> of the SAM model.</p></li>
 <li><p><strong>Markdown cells</strong> explaining your code, methodology, and results.</p></li>

outline.html

@@ -476,7 +476,8 @@ <h2>Weeks 2–3: Free recall, Short Term and Long Term Memory<a class="headerlin
 </li>
 <li><p>Readings:</p>
 <ul>
-<li><p><a class="reference external" href="https://www.dropbox.com/scl/fi/rpllozjcv704okckjdy5k/AtkiShif68.pdf?rlkey=i0azhj9mqxws7bxocbl65j88d">Atkinson and Shiffrin (1968)</a>*</p></li>
+<li><p><a class="reference external" href="https://www.dropbox.com/scl/fi/rpllozjcv704okckjdy5k/AtkiShif68.pdf?rlkey=i0azhj9mqxws7bxocbl65j88d">Atkinson and Shiffrin (1968)</a></p></li>
+<li><p><a class="reference external" href="https://www.dropbox.com/scl/fi/ujl8yvxqzcb1gf32to4zb/Kaha12_SAM_model_excerpt.pdf?rlkey=254wtw4fm7xnpzelno2ykrxzu">Kahana (2012), excerpt from Chapter 7</a>*</p></li>
 <li><p><a class="reference external" href="https://www.dropbox.com/scl/fi/wg6fledn7g88ig5mk3kob/ChenEtal16.pdf?rlkey=9jqu7y2apqv2hrj8qepn4alwa">Chen et al. (2016)</a></p></li>
 <li><p><a class="reference external" href="https://www.dropbox.com/scl/fi/w7z2yvdfzmhowh5hvg53e/HeusEtal21.pdf?rlkey=omad9klqeiu2kc71w7guc5xxq">Heusser et al. (2021)</a></p></li>
 </ul>